REINDEER: A Protein-Ligand Feature Generator Software for Machine Learning Algorithms

Machine learning-based scoring functions, which apply feature-generation methods for protein-ligand representation, have become ubiquitous in the past few years for binding affinity prediction. However, most of these feature-generation techniques are hidden inside papers and their corresponding codes. In this manuscript, we introduced REINDEER software to make these methods accessible to other users. REINDEER has been developed based on minimum dependencies and parallelization aims by Python programming language. The current version of REINDEER (v0.1.0) only includes feature generation methods from RF-Score, ET-Score, ECIF∷LD-GBT, and OnionNet-2 scoring functions. REINDEER provides a command line interface, graphical user interface, and usage within Python code capabilities to access these methods. Also, a case study on PDBbind refined set v2020 is presented to evaluate REINDEER abilities. REINDEER software is available at https://github.com/miladrayka/reindeer_software.